Neuromodulation Using Optogenetics and Related Technologies

Optogenetic Effector Delivery

Typically, viral vectors are used, but transgenic techniques and the direct introduction of genetic material (e.g., electroporation) have been used in basic science applications. Most commonly, adeno-associated viral (AAV) vectors are used and are readily available (e.g., University of North Carolina Vector Core Facility) and include a variety of promoters or in a FLEX configuration ( ), so that expression is limited to Cre recombinase–expressing neurons. Cre recombinase and loxP constitute a bacteriophage recombination system, with the Cre recombinase enzyme recombining nucleotides flanking loxP restriction sites. This has been used in mammals and neuroscience research to knock-out genes of interest at a chosen age and location within the nervous system ( ). These latter approaches afford cell-type specificity, with anatomical specificity that is determined by stereotactic microinjection. Expression is often adequate at about 4 weeks after injection, with optical fiber insertion at the time of stereotactic injection preferred to facilitate anatomical colocalization of the transduced protein and light source. In the primate brain, vector delivery needs to include sufficient tissue for the desired effect as well as consideration of vector tropism (the tendency for the vector to be taken up or expressed more avidly in particular cell types), with active work on these issues under way ( ).

Light Delivery, Tissue Penetration, and Scattering

Light delivery in basic science applications typically relies on laser diodes and flexible optical fibers ( ), but activation of a large volume of tissue will likely require the development of alternate light delivery strategies, including flexible light emitting diode (LED) arrays ( ), waveguides ( ), or diffusing fibers ( ). If more than one site of light delivery is required, then multiple light sources or beam splitting is required. Light sources consist of either an LED or a laser source, with LEDs now becoming the most common and convenient, given the potential for miniaturization and both low-power consumption and minimal heating. The LED and laser are controlled by a current source and train generator that together determine light intensity, duration of illumination, frequency, and stimulus train characteristics. The onset and offset of maximal illumination must also be considered, given the delay in maximal illumination when some LEDs or lasers are turned on and off, with temporal precision improved by the use of a shutter to occlude or reveal the light source.

While light intensity from a point source falls exponentially, there are additional considerations when illuminating brain tissue. The degree of lensing of light exiting the fiber is an obvious consideration that can be deliberately manipulated. Light scattering varies by brain region, depending on tissue optical properties, with variation as much as 1 order of magnitude between different brain regions. Finally, the wavelength of light has a significant effect on the depth of penetrance, with longer wavelengths significantly less susceptible to scattering and therefore having deeper tissue penetration ( ). However, while longer wavelengths penetrate tissue to a greater distance, tissue heating is increased ( ).

Toxicity and Nonoptogenetic Effects

There are four general untoward effects of stimulation that need to be considered: direct phototoxicity, toxicity of an excess of ion movement or exhaustion, the toxic effects of protein overexpression, and heating of tissue. While some have suggested that continued ability to activate or inhibit tissue is suggestive of the absence of toxicity ( ), it is possible that physiologic effects are due to cells outside of the area of potential toxicity that is immediately adjacent to the probe. Given the need for long-term stable stimulation, histologic verification of the absence of phototoxicity, in an appropriate, nonhuman model seems desirable. Phototoxicity is widely observed and appreciated in the setting of in vitro slice and cell culture work that requires fluorescent illumination but seems to have been neglected in in vivo optogenetic studies. Illumination below 10–20 mW/mm ² (depending on wavelength) is likely safe and does not result in signs of photodamage ( ) – the temptation to increase illumination intensity to improve effect size and tissue penetration in larger brains needs to be avoided. While direct phototoxicity is likely related to the generation of oxidative radicals, activation-related toxicity may also include large ionic or pH shifts that are potentially metabolically overwhelming. The amount of expression of channel is difficult to control (requires optimization of delivery and vector, consideration of viral vector tropism), so these large ionic fluxes, produced by a channel that is not normally expressed, particularly in large amounts, represent one motivation for using alternative techniques such as d esigner r eceptors e xclusively a ctivated by d esigner d rugs (DREADDs, also referred to as “chemogenetics” or “pharmacogenetics,” see later) that rely on natively expressed G protein–coupled mechanisms. Overexpression of a single introduced protein or use of very high vector titers can be cytotoxic, which in turn depend on vector serotype and cellular tropism ( ). Adequate examination for cytotoxicity is presently not often performed in basic science applications. Finally, tissue temperature changes need to be considered, given time-dependent tissue damage above 44–45°C and thermal coagulation at 60°C. Lesser increases in temperature tend to plateau during a given intensity and duty cycle of stimulation but result in changes in neural activity – a significant, potential confound in preclinical studies ( ). In the case of translation, nonhuman primate or larger animal studies could be considered to further verify the absence of accumulating phototoxic effects, toxic effects of protein overexpression, and heating effects.

Duty Cycle/Stimulation Parameters

Ideally, one needs to choose an optogenetic protein that will permit the stimulation/inhibition characteristics that are desired while avoiding the toxic and heating effects of stimulation. This leads to the consideration of both the type of optogenetic protein used and the stimulation parameters used. The kinetics of optogenetic protein state changes determines the maximal reasonable stimulation frequency. For example, channelrhodopsin-2 (ChR2), owing to channel closing time, typically will not enable light-trigger neuronal firing (i.e., cannot “follow”) above frequencies of about 20 Hz ( ). In contrast, the modified ChR2 variant, ChETA, will follow at up to at least 200 Hz and result in less doublet firing and less associated ion movement ( ), given more rapid channel inactivation. Similarly, excess activation of archaerhodopsin results in large proton shifts that result in a refractory period for cellular activity, and there can be significant rebound firing at the end of halorhodipsin activation. An excellent resource for the selection of an appropriate optogenetic tool is the Stanford Online Optogenetics resource as well as a recent text ( ). This selection needs to be combined with an understanding of the physiologic characteristics and desired manipulation of the target neuronal population.

Amenable Brain Targets

While optogenetic techniques have been very useful in rodent studies, the range of scientific questions and ease of use in therapies for experimental animals and humans with larger brains are limited by scale. Only a small area can be illuminated (1 mm ³ ) from existing light-delivery devices. Nontoxic levels of illumination, such as 3 mW/mm ² , result in limited tissue heating but illuminate only about 1–1.5 mm of tissue immediately surrounding the optical fiber, depending on the wavelength ( ). Nuclei of this size that would be an effective functional neurosurgical target are few, with possible exceptions of the basal forebrain (improving level of arousal, attention), nucleus accumbens (for obsessive compulsive disorder and addictions), amygdala subnuclei (including treatment of anxiety), thalamic subnuclei (for potential seizure termination, indirect cortical modulation, tremor) and other diencephalic targets, including the approximately 5-mm subthalamic nucleus (treatment of Parkinson disease [PD]). Even for these smaller human brain targets, a conventional single fiber seems unlikely to provide adequate coverage (see earlier for alternatives). Moving beyond available technologies, while one can imagine a diffuser device or array for the cortical surface, this would not easily access sulci, given significant vasculature in the sulcus (and thus only covers about the one third of cortex at the gyral crest), or complicated subcortical targets, such as the cerebellum. Understanding the involvement of smaller targets in human functional disorders will aid in the employment of optogenetic vector technologies for treatment, but obtaining safe stable and not excessive construct expression remains an important consideration.

Rebound and Unexpected Effects of Opsins

In addition to these potential toxic effects of this technology are unexpected physiologic effects, including a dissociation between neuronal firing and transmitter release, as well as rebound firing. Interestingly, transmitter release has been shown to increase after activation of putatively inhibitory opsins that are expressed in presynaptic terminals, including increased release on light onset with a modified channelrhodopsin chloride channel and increased transmitter release after sustained archaerhodopsin activation ( ). Rebound phenomena are widely appreciated with halorhodopsin and archaerhodopsin but can be attenuated by gradually tapering light offset ( ), while determination of this effect in vivo requires simultaneous and often technically difficult measurement of firing of manipulated neurons.

Light-Sensitive Transporters

The major opsin transporter proteins, archaerhodopsin and halorhodopsin, derive from the halophilic bacteria, Halobacterium sp. and Natronomonas pharaonis ( ).